/src/icu/icu4c/source/i18n/number_affixutils.cpp

Source
// © 2017 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html

#include "unicode/utypes.h"

#if !UCONFIG_NO_FORMATTING

#include "number_affixutils.h"
#include "unicode/utf16.h"
#include "unicode/uniset.h"

using namespace icu;
using namespace icu::number;
using namespace icu::number::impl;

TokenConsumer::~TokenConsumer() = default;
SymbolProvider::~SymbolProvider() = default;

int32_t AffixUtils::estimateLength(const UnicodeString &patternString, UErrorCode &status) {
    AffixPatternState state = STATE_BASE;
    int32_t offset = 0;
    int32_t length = 0;
    for (; offset < patternString.length();) {
        UChar32 cp = patternString.char32At(offset);

        switch (state) {
            case STATE_BASE:
                if (cp == u'\'') {
                    // First quote
                    state = STATE_FIRST_QUOTE;
                } else {
                    // Unquoted symbol
                    length++;
                }
                break;
            case STATE_FIRST_QUOTE:
                if (cp == u'\'') {
                    // Repeated quote
                    length++;
                    state = STATE_BASE;
                } else {
                    // Quoted code point
                    length++;
                    state = STATE_INSIDE_QUOTE;
                }
                break;
            case STATE_INSIDE_QUOTE:
                if (cp == u'\'') {
                    // End of quoted sequence
                    state = STATE_AFTER_QUOTE;
                } else {
                    // Quoted code point
                    length++;
                }
                break;
            case STATE_AFTER_QUOTE:
                if (cp == u'\'') {
                    // Double quote inside of quoted sequence
                    length++;
                    state = STATE_INSIDE_QUOTE;
                } else {
                    // Unquoted symbol
                    length++;
                }
                break;
            default:
                UPRV_UNREACHABLE_EXIT;
        }

        offset += U16_LENGTH(cp);
    }

    switch (state) {
        case STATE_FIRST_QUOTE:
        case STATE_INSIDE_QUOTE:
            status = U_ILLEGAL_ARGUMENT_ERROR;
            break;
        default:
            break;
    }

    return length;
}

UnicodeString AffixUtils::escape(const UnicodeString &input) {
    AffixPatternState state = STATE_BASE;
    int32_t offset = 0;
    UnicodeString output;
    for (; offset < input.length();) {
        UChar32 cp = input.char32At(offset);

        switch (cp) {
            case u'\'':
                output.append(u"''", -1);
                break;

            case u'-':
            case u'+':
            case u'%':
            case u'‰':
            case u'¤':
                if (state == STATE_BASE) {
                    output.append(u'\'');
                    output.append(cp);
                    state = STATE_INSIDE_QUOTE;
                } else {
                    output.append(cp);
                }
                break;

            default:
                if (state == STATE_INSIDE_QUOTE) {
                    output.append(u'\'');
                    output.append(cp);
                    state = STATE_BASE;
                } else {
                    output.append(cp);
                }
                break;
        }
        offset += U16_LENGTH(cp);
    }

    if (state == STATE_INSIDE_QUOTE) {
        output.append(u'\'');
    }

    return output;
}

Field AffixUtils::getFieldForType(AffixPatternType type) {
    switch (type) {
        case TYPE_MINUS_SIGN:
            return {UFIELD_CATEGORY_NUMBER, UNUM_SIGN_FIELD};
        case TYPE_PLUS_SIGN:
            return {UFIELD_CATEGORY_NUMBER, UNUM_SIGN_FIELD};
        case TYPE_APPROXIMATELY_SIGN:
            return {UFIELD_CATEGORY_NUMBER, UNUM_APPROXIMATELY_SIGN_FIELD};
        case TYPE_PERCENT:
            return {UFIELD_CATEGORY_NUMBER, UNUM_PERCENT_FIELD};
        case TYPE_PERMILLE:
            return {UFIELD_CATEGORY_NUMBER, UNUM_PERMILL_FIELD};
        case TYPE_CURRENCY_SINGLE:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_DOUBLE:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_TRIPLE:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_QUAD:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_QUINT:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        case TYPE_CURRENCY_OVERFLOW:
            return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
        default:
            UPRV_UNREACHABLE_EXIT;
    }
}

int32_t
AffixUtils::unescape(const UnicodeString &affixPattern, FormattedStringBuilder &output, int32_t position,
                     const SymbolProvider &provider, Field field, UErrorCode &status) {
    int32_t length = 0;
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return length; }
        if (tag.type == TYPE_CURRENCY_OVERFLOW) {
            // Don't go to the provider for this special case
            length += output.insertCodePoint(
                position + length,
                0xFFFD,
                {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD},
                status);
        } else if (tag.type < 0) {
            length += output.insert(
                    position + length, provider.getSymbol(tag.type), getFieldForType(tag.type), status);
        } else {
            length += output.insertCodePoint(position + length, tag.codePoint, field, status);
        }
    }
    return length;
}

int32_t AffixUtils::unescapedCodePointCount(const UnicodeString &affixPattern,
                                            const SymbolProvider &provider, UErrorCode &status) {
    int32_t length = 0;
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return length; }
        if (tag.type == TYPE_CURRENCY_OVERFLOW) {
            length += 1;
        } else if (tag.type < 0) {
            length += provider.getSymbol(tag.type).length();
        } else {
            length += U16_LENGTH(tag.codePoint);
        }
    }
    return length;
}

bool
AffixUtils::containsType(const UnicodeString &affixPattern, AffixPatternType type, UErrorCode &status) {
    if (affixPattern.length() == 0) {
        return false;
    }
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return false; }
        if (tag.type == type) {
            return true;
        }
    }
    return false;
}

bool AffixUtils::hasCurrencySymbols(const UnicodeString &affixPattern, UErrorCode &status) {
    if (affixPattern.length() == 0) {
        return false;
    }
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return false; }
        if (tag.type < 0 && getFieldForType(tag.type) == Field(UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD)) {
            return true;
        }
    }
    return false;
}

UnicodeString AffixUtils::replaceType(const UnicodeString &affixPattern, AffixPatternType type,
                                      char16_t replacementChar, UErrorCode &status) {
    UnicodeString output(affixPattern); // copy
    if (affixPattern.length() == 0) {
        return output;
    }
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return output; }
        if (tag.type == type) {
            output.replace(tag.offset - 1, 1, replacementChar);
        }
    }
    return output;
}

bool AffixUtils::containsOnlySymbolsAndIgnorables(const UnicodeString& affixPattern,
                                                  const UnicodeSet& ignorables, UErrorCode& status) {
    if (affixPattern.length() == 0) {
        return true;
    }
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return false; }
        if (tag.type == TYPE_CODEPOINT && !ignorables.contains(tag.codePoint)) {
            return false;
        }
    }
    return true;
}

void AffixUtils::iterateWithConsumer(const UnicodeString& affixPattern, TokenConsumer& consumer,
                                     UErrorCode& status) {
    if (affixPattern.length() == 0) {
        return;
    }
    AffixTag tag;
    while (hasNext(tag, affixPattern)) {
        tag = nextToken(tag, affixPattern, status);
        if (U_FAILURE(status)) { return; }
        consumer.consumeToken(tag.type, tag.codePoint, status);
        if (U_FAILURE(status)) { return; }
    }
}

AffixTag AffixUtils::nextToken(AffixTag tag, const UnicodeString &patternString, UErrorCode &status) {
    int32_t offset = tag.offset;
    int32_t state = tag.state;
    for (; offset < patternString.length();) {
        UChar32 cp = patternString.char32At(offset);
        int32_t count = U16_LENGTH(cp);

        switch (state) {
            case STATE_BASE:
                switch (cp) {
                    case u'\'':
                        state = STATE_FIRST_QUOTE;
                        offset += count;
                        // continue to the next code point
                        break;
                    case u'-':
                        return makeTag(offset + count, TYPE_MINUS_SIGN, STATE_BASE, 0);
                    case u'+':
                        return makeTag(offset + count, TYPE_PLUS_SIGN, STATE_BASE, 0);
                    case u'~':
                        return makeTag(offset + count, TYPE_APPROXIMATELY_SIGN, STATE_BASE, 0);
                    case u'%':
                        return makeTag(offset + count, TYPE_PERCENT, STATE_BASE, 0);
                    case u'‰':
                        return makeTag(offset + count, TYPE_PERMILLE, STATE_BASE, 0);
                    case u'¤':
                        state = STATE_FIRST_CURR;
                        offset += count;
                        // continue to the next code point
                        break;
                    default:
                        return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
                }
                break;
            case STATE_FIRST_QUOTE:
                if (cp == u'\'') {
                    return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
                } else {
                    return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
                }
            case STATE_INSIDE_QUOTE:
                if (cp == u'\'') {
                    state = STATE_AFTER_QUOTE;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
                }
            case STATE_AFTER_QUOTE:
                if (cp == u'\'') {
                    return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
                } else {
                    state = STATE_BASE;
                    // re-evaluate this code point
                    break;
                }
            case STATE_FIRST_CURR:
                if (cp == u'¤') {
                    state = STATE_SECOND_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
                }
            case STATE_SECOND_CURR:
                if (cp == u'¤') {
                    state = STATE_THIRD_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
                }
            case STATE_THIRD_CURR:
                if (cp == u'¤') {
                    state = STATE_FOURTH_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
                }
            case STATE_FOURTH_CURR:
                if (cp == u'¤') {
                    state = STATE_FIFTH_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
                }
            case STATE_FIFTH_CURR:
                if (cp == u'¤') {
                    state = STATE_OVERFLOW_CURR;
                    offset += count;
                    // continue to the next code point
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
                }
            case STATE_OVERFLOW_CURR:
                if (cp == u'¤') {
                    offset += count;
                    // continue to the next code point and loop back to this state
                    break;
                } else {
                    return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
                }
            default:
                UPRV_UNREACHABLE_EXIT;
        }
    }
    // End of string
    switch (state) {
        case STATE_BASE:
            // No more tokens in string.
            return {-1};
        case STATE_FIRST_QUOTE:
        case STATE_INSIDE_QUOTE:
            // For consistent behavior with the JDK and ICU 58, set an error here.
            status = U_ILLEGAL_ARGUMENT_ERROR;
            return {-1};
        case STATE_AFTER_QUOTE:
            // No more tokens in string.
            return {-1};
        case STATE_FIRST_CURR:
            return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
        case STATE_SECOND_CURR:
            return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
        case STATE_THIRD_CURR:
            return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
        case STATE_FOURTH_CURR:
            return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
        case STATE_FIFTH_CURR:
            return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
        case STATE_OVERFLOW_CURR:
            return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
        default:
            UPRV_UNREACHABLE_EXIT;
    }
}

bool AffixUtils::hasNext(const AffixTag &tag, const UnicodeString &string) {
    // First check for the {-1} and default initializer syntax.
    if (tag.offset < 0) {
        return false;
    } else if (tag.offset == 0) {
        return string.length() > 0;
    }
    // The rest of the fields are safe to use now.
    // Special case: the last character in string is an end quote.
    if (tag.state == STATE_INSIDE_QUOTE && tag.offset == string.length() - 1 &&
        string.charAt(tag.offset) == u'\'') {
        return false;
    } else if (tag.state != STATE_BASE) {
        return true;
    } else {
        return tag.offset < string.length();
    }
}

#endif /* #if !UCONFIG_NO_FORMATTING */

Coverage Report

Created: 2025-12-07 06:36

Line	Count	Source
1		// © 2017 and later: Unicode, Inc. and others.
2		// License & terms of use: http://www.unicode.org/copyright.html
3
4		#include "unicode/utypes.h"
5
6		#if !UCONFIG_NO_FORMATTING
7
8		#include "number_affixutils.h"
9		#include "unicode/utf16.h"
10		#include "unicode/uniset.h"
11
12		using namespace icu;
13		using namespace icu::number;
14		using namespace icu::number::impl;
15
16	171k	TokenConsumer::~TokenConsumer() = default;
17	98.9k	SymbolProvider::~SymbolProvider() = default;
18
19	7.45k	int32_t AffixUtils::estimateLength(const UnicodeString &patternString, UErrorCode &status) {
20	7.45k	AffixPatternState state = STATE_BASE;
21	7.45k	int32_t offset = 0;
22	7.45k	int32_t length = 0;
23	3.90M	for (; offset < patternString.length();) {
24	3.89M	UChar32 cp = patternString.char32At(offset);
25
26	3.89M	switch (state) {
27	3.80M	case STATE_BASE:
28	3.80M	if (cp == u'\'') {
29		// First quote
30	433	state = STATE_FIRST_QUOTE;
31	3.80M	} else {
32		// Unquoted symbol
33	3.80M	length++;
34	3.80M	}
35	3.80M	break;
36	433	case STATE_FIRST_QUOTE:
37	433	if (cp == u'\'') {
38		// Repeated quote
39	341	length++;
40	341	state = STATE_BASE;
41	341	} else {
42		// Quoted code point
43	92	length++;
44	92	state = STATE_INSIDE_QUOTE;
45	92	}
46	433	break;
47	66.1k	case STATE_INSIDE_QUOTE:
48	66.1k	if (cp == u'\'') {
49		// End of quoted sequence
50	5.55k	state = STATE_AFTER_QUOTE;
51	60.6k	} else {
52		// Quoted code point
53	60.6k	length++;
54	60.6k	}
55	66.1k	break;
56	23.3k	case STATE_AFTER_QUOTE:
57	23.3k	if (cp == u'\'') {
58		// Double quote inside of quoted sequence
59	5.46k	length++;
60	5.46k	state = STATE_INSIDE_QUOTE;
61	17.8k	} else {
62		// Unquoted symbol
63	17.8k	length++;
64	17.8k	}
65	23.3k	break;
66	0	default:
67	0	UPRV_UNREACHABLE_EXIT;
68	3.89M	}
69
70	3.89M	offset += U16_LENGTH(cp);
71	3.89M	}
72
73	7.45k	switch (state) {
74	0	case STATE_FIRST_QUOTE:
75	0	case STATE_INSIDE_QUOTE:
76	0	status = U_ILLEGAL_ARGUMENT_ERROR;
77	0	break;
78	7.45k	default:
79	7.45k	break;
80	7.45k	}
81
82	7.45k	return length;
83	7.45k	}
84
85	2.55M	UnicodeString AffixUtils::escape(const UnicodeString &input) {
86	2.55M	AffixPatternState state = STATE_BASE;
87	2.55M	int32_t offset = 0;
88	2.55M	UnicodeString output;
89	2.55M	for (; offset < input.length();) {
90	0	UChar32 cp = input.char32At(offset);
91
92	0	switch (cp) {
93	0	case u'\'':
94	0	output.append(u"''", -1);
95	0	break;
96
97	0	case u'-':
98	0	case u'+':
99	0	case u'%':
100	0	case u'‰':
101	0	case u'¤':
102	0	if (state == STATE_BASE) {
103	0	output.append(u'\'');
104	0	output.append(cp);
105	0	state = STATE_INSIDE_QUOTE;
106	0	} else {
107	0	output.append(cp);
108	0	}
109	0	break;
110
111	0	default:
112	0	if (state == STATE_INSIDE_QUOTE) {
113	0	output.append(u'\'');
114	0	output.append(cp);
115	0	state = STATE_BASE;
116	0	} else {
117	0	output.append(cp);
118	0	}
119	0	break;
120	0	}
121	0	offset += U16_LENGTH(cp);
122	0	}
123
124	2.55M	if (state == STATE_INSIDE_QUOTE) {
125	0	output.append(u'\'');
126	0	}
127
128	2.55M	return output;
129	2.55M	}
130
131	310k	Field AffixUtils::getFieldForType(AffixPatternType type) {
132	310k	switch (type) {
133	97.1k	case TYPE_MINUS_SIGN:
134	97.1k	return {UFIELD_CATEGORY_NUMBER, UNUM_SIGN_FIELD};
135	93.4k	case TYPE_PLUS_SIGN:
136	93.4k	return {UFIELD_CATEGORY_NUMBER, UNUM_SIGN_FIELD};
137	891	case TYPE_APPROXIMATELY_SIGN:
138	891	return {UFIELD_CATEGORY_NUMBER, UNUM_APPROXIMATELY_SIGN_FIELD};
139	27.2k	case TYPE_PERCENT:
140	27.2k	return {UFIELD_CATEGORY_NUMBER, UNUM_PERCENT_FIELD};
141	51.8k	case TYPE_PERMILLE:
142	51.8k	return {UFIELD_CATEGORY_NUMBER, UNUM_PERMILL_FIELD};
143	38.2k	case TYPE_CURRENCY_SINGLE:
144	38.2k	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
145	505	case TYPE_CURRENCY_DOUBLE:
146	505	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
147	967	case TYPE_CURRENCY_TRIPLE:
148	967	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
149	194	case TYPE_CURRENCY_QUAD:
150	194	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
151	80	case TYPE_CURRENCY_QUINT:
152	80	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
153	65	case TYPE_CURRENCY_OVERFLOW:
154	65	return {UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD};
155	0	default:
156	0	UPRV_UNREACHABLE_EXIT;
157	310k	}
158	310k	}
159
160		int32_t
161		AffixUtils::unescape(const UnicodeString &affixPattern, FormattedStringBuilder &output, int32_t position,
162	732k	const SymbolProvider &provider, Field field, UErrorCode &status) {
163	732k	int32_t length = 0;
164	732k	AffixTag tag;
165	921k	while (hasNext(tag, affixPattern)) {
166	188k	tag = nextToken(tag, affixPattern, status);
167	188k	if (U_FAILURE(status)) { return length; }
168	188k	if (tag.type == TYPE_CURRENCY_OVERFLOW) {
169		// Don't go to the provider for this special case
170	0	length += output.insertCodePoint(
171	0	position + length,
172	0	0xFFFD,
173	0	{UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD},
174	0	status);
175	188k	} else if (tag.type < 0) {
176	180k	length += output.insert(
177	180k	position + length, provider.getSymbol(tag.type), getFieldForType(tag.type), status);
178	180k	} else {
179	7.14k	length += output.insertCodePoint(position + length, tag.codePoint, field, status);
180	7.14k	}
181	188k	}
182	732k	return length;
183	732k	}
184
185		int32_t AffixUtils::unescapedCodePointCount(const UnicodeString &affixPattern,
186	0	const SymbolProvider &provider, UErrorCode &status) {
187	0	int32_t length = 0;
188	0	AffixTag tag;
189	0	while (hasNext(tag, affixPattern)) {
190	0	tag = nextToken(tag, affixPattern, status);
191	0	if (U_FAILURE(status)) { return length; }
192	0	if (tag.type == TYPE_CURRENCY_OVERFLOW) {
193	0	length += 1;
194	0	} else if (tag.type < 0) {
195	0	length += provider.getSymbol(tag.type).length();
196	0	} else {
197	0	length += U16_LENGTH(tag.codePoint);
198	0	}
199	0	}
200	0	return length;
201	0	}
202
203		bool
204	36.7k	AffixUtils::containsType(const UnicodeString &affixPattern, AffixPatternType type, UErrorCode &status) {
205	36.7k	if (affixPattern.length() == 0) {
206	13.2k	return false;
207	13.2k	}
208	23.5k	AffixTag tag;
209	8.69M	while (hasNext(tag, affixPattern)) {
210	8.67M	tag = nextToken(tag, affixPattern, status);
211	8.67M	if (U_FAILURE(status)) { return false; }
212	8.67M	if (tag.type == type) {
213	2.63k	return true;
214	2.63k	}
215	8.67M	}
216	20.9k	return false;
217	23.5k	}
218
219	2.46M	bool AffixUtils::hasCurrencySymbols(const UnicodeString &affixPattern, UErrorCode &status) {
220	2.46M	if (affixPattern.length() == 0) {
221	2.34M	return false;
222	2.34M	}
223	113k	AffixTag tag;
224	69.1M	while (hasNext(tag, affixPattern)) {
225	69.0M	tag = nextToken(tag, affixPattern, status);
226	69.0M	if (U_FAILURE(status)) { return false; }
227	69.0M	if (tag.type < 0 && getFieldForType(tag.type) == Field(UFIELD_CATEGORY_NUMBER, UNUM_CURRENCY_FIELD)) {
228	40.0k	return true;
229	40.0k	}
230	69.0M	}
231	73.1k	return false;
232	113k	}
233
234		UnicodeString AffixUtils::replaceType(const UnicodeString &affixPattern, AffixPatternType type,
235	0	char16_t replacementChar, UErrorCode &status) {
236	0	UnicodeString output(affixPattern); // copy
237	0	if (affixPattern.length() == 0) {
238	0	return output;
239	0	}
240	0	AffixTag tag;
241	0	while (hasNext(tag, affixPattern)) {
242	0	tag = nextToken(tag, affixPattern, status);
243	0	if (U_FAILURE(status)) { return output; }
244	0	if (tag.type == type) {
245	0	output.replace(tag.offset - 1, 1, replacementChar);
246	0	}
247	0	}
248	0	return output;
249	0	}
250
251		bool AffixUtils::containsOnlySymbolsAndIgnorables(const UnicodeString& affixPattern,
252	0	const UnicodeSet& ignorables, UErrorCode& status) {
253	0	if (affixPattern.length() == 0) {
254	0	return true;
255	0	}
256	0	AffixTag tag;
257	0	while (hasNext(tag, affixPattern)) {
258	0	tag = nextToken(tag, affixPattern, status);
259	0	if (U_FAILURE(status)) { return false; }
260	0	if (tag.type == TYPE_CODEPOINT && !ignorables.contains(tag.codePoint)) {
261	0	return false;
262	0	}
263	0	}
264	0	return true;
265	0	}
266
267		void AffixUtils::iterateWithConsumer(const UnicodeString& affixPattern, TokenConsumer& consumer,
268	171k	UErrorCode& status) {
269	171k	if (affixPattern.length() == 0) {
270	0	return;
271	0	}
272	171k	AffixTag tag;
273	66.0M	while (hasNext(tag, affixPattern)) {
274	65.8M	tag = nextToken(tag, affixPattern, status);
275	65.8M	if (U_FAILURE(status)) { return; }
276	65.8M	consumer.consumeToken(tag.type, tag.codePoint, status);
277	65.8M	if (U_FAILURE(status)) { return; }
278	65.8M	}
279	171k	}
280
281	143M	AffixTag AffixUtils::nextToken(AffixTag tag, const UnicodeString &patternString, UErrorCode &status) {
282	143M	int32_t offset = tag.offset;
283	143M	int32_t state = tag.state;
284	144M	for (; offset < patternString.length();) {
285	144M	UChar32 cp = patternString.char32At(offset);
286	144M	int32_t count = U16_LENGTH(cp);
287
288	144M	switch (state) {
289	139M	case STATE_BASE:
290	139M	switch (cp) {
291	310k	case u'\'':
292	310k	state = STATE_FIRST_QUOTE;
293	310k	offset += count;
294		// continue to the next code point
295	310k	break;
296	206k	case u'-':
297	206k	return makeTag(offset + count, TYPE_MINUS_SIGN, STATE_BASE, 0);
298	98.6k	case u'+':
299	98.6k	return makeTag(offset + count, TYPE_PLUS_SIGN, STATE_BASE, 0);
300	1.72k	case u'~':
301	1.72k	return makeTag(offset + count, TYPE_APPROXIMATELY_SIGN, STATE_BASE, 0);
302	56.1k	case u'%':
303	56.1k	return makeTag(offset + count, TYPE_PERCENT, STATE_BASE, 0);
304	102k	case u'‰':
305	102k	return makeTag(offset + count, TYPE_PERMILLE, STATE_BASE, 0);
306	101k	case u'¤':
307	101k	state = STATE_FIRST_CURR;
308	101k	offset += count;
309		// continue to the next code point
310	101k	break;
311	138M	default:
312	138M	return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
313	139M	}
314	411k	break;
315	411k	case STATE_FIRST_QUOTE:
316	310k	if (cp == u'\'') {
317	260k	return makeTag(offset + count, TYPE_CODEPOINT, STATE_BASE, cp);
318	260k	} else {
319	49.4k	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
320	49.4k	}
321	4.25M	case STATE_INSIDE_QUOTE:
322	4.25M	if (cp == u'\'') {
323	309k	state = STATE_AFTER_QUOTE;
324	309k	offset += count;
325		// continue to the next code point
326	309k	break;
327	3.94M	} else {
328	3.94M	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
329	3.94M	}
330	309k	case STATE_AFTER_QUOTE:
331	309k	if (cp == u'\'') {
332	261k	return makeTag(offset + count, TYPE_CODEPOINT, STATE_INSIDE_QUOTE, cp);
333	261k	} else {
334	48.4k	state = STATE_BASE;
335		// re-evaluate this code point
336	48.4k	break;
337	48.4k	}
338	43.7k	case STATE_FIRST_CURR:
339	43.7k	if (cp == u'¤') {
340	7.69k	state = STATE_SECOND_CURR;
341	7.69k	offset += count;
342		// continue to the next code point
343	7.69k	break;
344	36.0k	} else {
345	36.0k	return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
346	36.0k	}
347	7.10k	case STATE_SECOND_CURR:
348	7.10k	if (cp == u'¤') {
349	5.23k	state = STATE_THIRD_CURR;
350	5.23k	offset += count;
351		// continue to the next code point
352	5.23k	break;
353	5.23k	} else {
354	1.86k	return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
355	1.86k	}
356	3.94k	case STATE_THIRD_CURR:
357	3.94k	if (cp == u'¤') {
358	2.48k	state = STATE_FOURTH_CURR;
359	2.48k	offset += count;
360		// continue to the next code point
361	2.48k	break;
362	2.48k	} else {
363	1.45k	return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
364	1.45k	}
365	2.21k	case STATE_FOURTH_CURR:
366	2.21k	if (cp == u'¤') {
367	1.78k	state = STATE_FIFTH_CURR;
368	1.78k	offset += count;
369		// continue to the next code point
370	1.78k	break;
371	1.78k	} else {
372	430	return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
373	430	}
374	1.71k	case STATE_FIFTH_CURR:
375	1.71k	if (cp == u'¤') {
376	950	state = STATE_OVERFLOW_CURR;
377	950	offset += count;
378		// continue to the next code point
379	950	break;
380	950	} else {
381	760	return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
382	760	}
383	5.86k	case STATE_OVERFLOW_CURR:
384	5.86k	if (cp == u'¤') {
385	5.00k	offset += count;
386		// continue to the next code point and loop back to this state
387	5.00k	break;
388	5.00k	} else {
389	858	return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
390	858	}
391	0	default:
392	0	UPRV_UNREACHABLE_EXIT;
393	144M	}
394	144M	}
395		// End of string
396	59.5k	switch (state) {
397	0	case STATE_BASE:
398		// No more tokens in string.
399	0	return {-1};
400	0	case STATE_FIRST_QUOTE:
401	0	case STATE_INSIDE_QUOTE:
402		// For consistent behavior with the JDK and ICU 58, set an error here.
403	0	status = U_ILLEGAL_ARGUMENT_ERROR;
404	0	return {-1};
405	0	case STATE_AFTER_QUOTE:
406		// No more tokens in string.
407	0	return {-1};
408	57.2k	case STATE_FIRST_CURR:
409	57.2k	return makeTag(offset, TYPE_CURRENCY_SINGLE, STATE_BASE, 0);
410	591	case STATE_SECOND_CURR:
411	591	return makeTag(offset, TYPE_CURRENCY_DOUBLE, STATE_BASE, 0);
412	1.28k	case STATE_THIRD_CURR:
413	1.28k	return makeTag(offset, TYPE_CURRENCY_TRIPLE, STATE_BASE, 0);
414	272	case STATE_FOURTH_CURR:
415	272	return makeTag(offset, TYPE_CURRENCY_QUAD, STATE_BASE, 0);
416	75	case STATE_FIFTH_CURR:
417	75	return makeTag(offset, TYPE_CURRENCY_QUINT, STATE_BASE, 0);
418	92	case STATE_OVERFLOW_CURR:
419	92	return makeTag(offset, TYPE_CURRENCY_OVERFLOW, STATE_BASE, 0);
420	0	default:
421	0	UPRV_UNREACHABLE_EXIT;
422	59.5k	}
423	59.5k	}
424
425	144M	bool AffixUtils::hasNext(const AffixTag &tag, const UnicodeString &string) {
426		// First check for the {-1} and default initializer syntax.
427	144M	if (tag.offset < 0) {
428	0	return false;
429	144M	} else if (tag.offset == 0) {
430	1.04M	return string.length() > 0;
431	1.04M	}
432		// The rest of the fields are safe to use now.
433		// Special case: the last character in string is an end quote.
434	143M	if (tag.state == STATE_INSIDE_QUOTE && tag.offset == string.length() - 1 &&
435	1.06k	string.charAt(tag.offset) == u'\'') {
436	1.06k	return false;
437	143M	} else if (tag.state != STATE_BASE) {
438	4.25M	return true;
439	139M	} else {
440	139M	return tag.offset < string.length();
441	139M	}
442	143M	}
443
444		#endif /* #if !UCONFIG_NO_FORMATTING */